AI-Powered Job Tracker

A full-stack SaaS application built with the MERN stack that helps users manage job applications and leverages AI to analyze resume-job description matching.

Architecture

Client (React + Vite + Tailwind CSS)
        ↓
Express API (Node.js)
        ↓
MongoDB (Atlas or In-Memory)
        ↓
AI Provider (pick one)
├── Claude (Anthropic)
├── OpenAI (GPT)
├── Gemini (Google)
├── DeepSeek
└── Ollama (Local)

Features

• User Authentication — Register, login, and JWT-based session management
• Job Tracking — Create, read, update, and delete job applications
• Status Management — Track applications as Applied, Interview, Offer, or Rejected
• Dashboard Analytics — Visual stats summary with status-based filtering
• AI Resume Matching — Upload your resume (PDF or DOCX) to get an AI-powered match score and feedback against any job description
• Multi-Provider AI — Choose between Claude, OpenAI, Gemini, DeepSeek, or Ollama per analysis

Supported AI Providers

Provider	Model	API Key Required	Notes
Claude	claude-sonnet	Yes	Anthropic API
OpenAI	gpt-4o-mini	Yes	OpenAI API
Gemini	gemini-2.0-flash	Yes	Google AI API
DeepSeek	deepseek-chat	Yes	OpenAI-compatible API
Ollama	llama3 (default)	No	Runs locally, no cloud needed

Project Structure

ai-job-tracker/
├── client/                  # React Frontend
│   └── src/
│       ├── api/axios.js         # Axios instance with auth interceptor
│       ├── components/          # Navbar, JobCard, JobForm, ProtectedRoute
│       ├── context/             # AuthContext (login, register, logout)
│       └── pages/               # Login, Register, Dashboard, JobDetails
│
├── server/                  # Node + Express Backend
│   ├── config/db.js             # MongoDB connection (with in-memory fallback)
│   ├── controllers/             # Auth, Job, AI controllers
│   ├── middleware/              # JWT auth, error handling & file upload (multer)
│   ├── models/                  # User & Job Mongoose schemas
│   ├── routes/                  # Auth, Job, AI routes
│   ├── utils/aiService.js       # Multi-provider AI integration
│   ├── utils/extractText.js    # PDF & DOCX text extraction
│   └── server.js                # Express entry point
│
└── README.md

Getting Started

Prerequisites

• Node.js (v18+)
• MongoDB Atlas account, local MongoDB, or none (in-memory mode available for testing)
• At least one AI provider API key (or Ollama installed locally)

1. Clone and configure

git clone <your-repo-url>
cd ai-job-tracker

Copy the environment template and fill in your values:

cp server/.env.example server/.env

Note: If MONGO_URI is not set, the server automatically uses an in-memory MongoDB instance via mongodb-memory-server. This is useful for quick testing but data will not persist across server restarts.

PORT=5000
MONGO_URI=mongodb+srv://<username>:<password>@cluster.mongodb.net/ai-job-tracker  # Optional — omit for in-memory testing
JWT_SECRET=your_jwt_secret_key_here

# Add keys only for the providers you want to use
ANTHROPIC_API_KEY=your_anthropic_api_key_here
OPENAI_API_KEY=your_openai_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here
DEEPSEEK_API_KEY=your_deepseek_api_key_here

# Ollama (local — no API key needed, just have Ollama running)
OLLAMA_BASE_URL=http://localhost:11434
OLLAMA_MODEL=llama3

Variable	Required	Description
MONGO_URI	No	MongoDB connection string (Atlas or local). Falls back to in-memory DB if not set
JWT_SECRET	Yes	Any random string used to sign JWT tokens
ANTHROPIC_API_KEY	No	Only if using Claude provider
OPENAI_API_KEY	No	Only if using OpenAI provider
GEMINI_API_KEY	No	Only if using Gemini provider
DEEPSEEK_API_KEY	No	Only if using DeepSeek provider
OLLAMA_BASE_URL	No	Defaults to http://localhost:11434
OLLAMA_MODEL	No	Defaults to llama3

2. Install dependencies

cd server && npm install
cd ../client && npm install

3. Run the application

Start the backend and frontend in separate terminals:

# Terminal 1 — Backend (http://localhost:5000)
cd server && npm run dev

# Terminal 2 — Frontend (http://localhost:5173)
cd client && npm run dev

API Routes

Method	Endpoint	Description	Auth
POST	/api/auth/register	Register a new user	Public
POST	/api/auth/login	Login	Public
GET	/api/auth/profile	Get user profile	Required
GET	/api/jobs	Get all user jobs	Required
POST	/api/jobs	Create a job	Required
GET	/api/jobs/:id	Get a single job	Required
PUT	/api/jobs/:id	Update a job	Required
DELETE	/api/jobs/:id	Delete a job	Required
POST	/api/ai/analyze	AI resume analysis	Required

The /api/ai/analyze endpoint accepts a multipart/form-data request with a resume file (PDF or DOCX), a jobId field, and an optional provider field ("claude", "openai", "gemini", "deepseek", or "ollama"). Defaults to "claude" if not specified.

Tech Stack

Layer	Technology
Frontend	React, Vite, Tailwind CSS
Backend	Node.js, Express
Database	MongoDB (Atlas, local, or in-memory), Mongoose
Auth	JWT, bcryptjs
AI	Claude, OpenAI, Gemini, DeepSeek, Ollama
Upload	Multer (memory storage), pdf-parse, mammoth